| Laboratory is gathering the high-value equipment and instruments.The security issue of laboratory becomes the focus of the more and more manager.Traditional laboratory mange system like access control and video surveillance system are unable to undertake complex task due to it’s bad timeliness and single function.Developing an intelligence video surveillance system whit fast response and tracing back characteristics has not only theoretical significance but also practical value.The key features of the system in this paper including face recognition and intrusion detection, both of the two subsystems use IP camera as video source and were developed by Matlab software.The face recognition subsystem get the shot from the IP camera regularly. Cascade adaboost classifier locate the position of the face and the eyes, then do the angle correction based on the position of eyes, cut the face image and resize it to80*65pixel resolution, meanwhile, use the Tan&Triggs’s illumination normalization to weaken illumination effect and retain the information of the face maximum. The identify stage including Gabor filter, PC A, LDA and improved KNN, use PC A to do dimension reduction, use LDA to maximize the difference between classes, use improved KNN classifier to do the global optimization, then save the final results in the database and output the Attendance Rate of each person,. Intrusion detection subsystem was designed based on requirements. The subsystem could detect intrusion from irregular entry like window, sensing ambient temperature. The subsystem will send email to the managers if unexpected events were detected.The Matlab and IP camera based intelligence video surveillance system has the characteristics of easy development, implementation, control and expansion. It also has user-friendly interface and is easy to use. The automation and the security level can be improved by using this system.In order to improve the performance of system,some functions will be compiled to mex file. |
PDF Full Text Request